Interventional Photoacoustic Imaging

of the Human Placenta with Ultrasonic Trackingfor Minimally Invasive Fetal Surgeries

Wenfeng Xia1, Efthymios Maneas2, Daniil I. Nikitichev1, Charles A. Mosse1,Gustavo Sato dos Santos2, Tom Vercauteren2, Anna L. David3, Jan Deprest4,

Sebastien Ourselin2, Paul C. Beard1, and Adrien E. Desjardins1

1 Department of Medical Physics and Biomedical Engineering, University CollegeLondon, Gower Street, London WC1E 6BT, United Kingdom

2 Translational Imaging Group, Centre for Medical Image Computing,Department of Medical Physics and Biomedical Engineering, University College

London, Wolfson House, London WC1E 6BT, United Kingdom3 Institute for Women’s Health, University College London, 86-96 Chenies Mews,

London WC1E 6HX, United Kingdom4 Department of Obstetrics and Gynecology, University Hospitals KU Leuven,

Leuven, Belgiumwenfeng.xia@ucl.ac.uk

Abstract. Image guidance plays a central role in minimally invasivefetal surgery such as photocoagulation of inter-twin placental anasto-mosing vessels to treat twin-to-twin transfusion syndrome (TTTS). Fe-toscopic guidance provides insufficient sensitivity for imaging the vascu-lature that lies beneath the fetal placental surface due to strong lightscattering in biological tissues. Incomplete photocoagulation of anasta-moses is associated with postoperative complications and higher peri-natal mortality. In this study, we investigated the use of multi-spectralphotoacoustic (PA) imaging for better visualization of the placental vas-culature. Excitation light was delivered with an optical fiber with dimen-sions that are compatible with the working channel of a fetoscope. Imag-ing was performed on an ex vivo normal term human placenta collectedat Caesarean section birth. The photoacoustically-generated ultrasoundsignals were received by an external clinical linear array ultrasound imag-ing probe. A vein under illumination on the fetal placenta surface wasvisualized with PA imaging, and good correspondence was obtained be-tween the measured PA spectrum and the optical absorption spectrumof deoxygenated blood. The delivery fiber had an attached fiber opticultrasound sensor positioned directly adjacent to it, so that its spatialposition could be tracked by receiving transmissions from the ultrasoundimaging probe. This study provides strong indications that PA imagingin combination with ultrasonic tracking could be useful for detecting thehuman placental vasculature during minimally invasive fetal surgery.

c© Springer International Publishing Switzerland 2015N. Navab et al. (Eds.): MICCAI 2015, Part I, LNCS 9349, pp. 371–378, 2015.DOI: 10.1007/978-3-319-24553-9_46

372 W. Xia et al.

1 Introduction

Image guidance is a central component of minimally invasive fetal surgery fortreatment of twin-to-twin transfusion syndrome (TTTS). The gold standard fortreatment involves laser photocoagulation of anastamosing vessels on the fetalside of the placenta [1]. In current practice, anastomosing vessels are identifiedwith vessels along the equator using a fetoscope. Due to the limitations of thismodality, there is a risk that sub-surface vessels that are small and those thatare at the periphery of the placenta are missed, so that treatment is incomplete.Ultrasound (US) imaging with a probe positioned at the external surface of themother provides inadequate visualization for small placental vessels. Visualiza-tion with US can be particularly poor when the placenta is on the posterior partof the uterus, at a large distance from the imaging probe so that low US frequen-cies are required. Power Doppler (PD) was also proposed by several groups tovisualize the placenta vasculature [2,3]. One challenge with PD is that successfulvessel identification is strongly dependent on the skill of the operator and thevessel orientation; additionally, this technique lacks sensitivity for small vesselsdue to low US contrast for soft tissues.

Photoacoustic (PA) imaging has strong potential to provide guidance infor-mation during TTTS that is complementary to fetoscopy and external US imag-ing. With PA imaging, pulsed or temporally modulated excitation light is ab-sorbed in tissue, which causes temperature rises and subsequent generation of USwaves [4]. These US waves can be received with an imaging probe at the surfaceof a patient and reconstructed to form 2D or 3D images with contrast for tis-sue chromophores. Multispectral PA images, which are acquired by varying thewavelength of excitation light, can be used to provide quantitative informationabout chromophore concentrations [5].

Conventional implementations of PA imaging, in which excitation light isdelivered at the surface of the patient, may be suboptimal in the context ofTTTS. As placental vessels typically lie more than five centimeters below thesurface of the patient, the PA signals may be very low or undetectable due tothe prominence of scattering and absorption of excitation light [6]. In this study,light is delivered directly to the placental surface using an optical fiber that issufficiently small to be inserted into the instrument channel of a fetoscope. Thegenerated PA signals were detected by a commercial linear array US imagingprobe.

The PA imaging system in this study was extended to allow for real-timein-plane ultrasonic tracking of the optical fiber that delivers excitation light.Ultrasonic tracking was implemented with a fiber-optic hydrophone that waspositioned alongside the delivery fiber. Multispectral PA imaging and ultrasonictracking were performed on a human placenta ex vivo, and the measured pho-toacoustic spectrum from a vein was compared with the optical spectra of oxy-and deoxy-hemoglobin.

Interventional Photoacoustic Imaging of the Human Placenta 373

&

Fig. 1. (a) Schematic illustration of the photoacoustic imaging and ultrasonic trackingsystem. (b) Photograph of the fetal surface of a human placenta under the imagingprobe.

2 Materials and Methods

2.1 PA Imaging and Ultrasonic Tracking System

The system that was used for PA imaging and tracking was based on a clinical USimaging system with a 128-element, 10 MHz linear-array US probe (SonixMDP,Analogic Ultrasound, Richmond, BC, Canada) as shown in Figure 1a. It wasoperated in research mode, which provides the access to low-level libraries foracquisition of B-mode images, transmission of US pulses for tracking, and col-lection of pre-beamformed RF data through a 128 channel DAQ system (Sonix-DAQ, Analogic Ultrasound, Richmond, BC, Canada) for PA imaging.

PA Imaging. Pulsed light with multiple wavelengths was provided by an opti-cal parametric oscillator (OPO) system (VersaScan L-532, GWU-Lasertechnik,Erftstadt, Germany) pumped by a Nd:YAG laser (pulse width 6 ns, repetitionrate 10 Hz, Quanta-Ray, INDI-40-10, Spectra-Physics, Santa Clara, CA). Thesignal and the idler from the OPO provided two wavelength ranges: 700-900 nmand 1100-2200 nm respectively. In this study, only the signal output was used(Figure 1a). The OPO output was coupled into an optical fiber with a 910 µmcore diameter (FG910LEC, Thorlabs, Newton, NJ). A small portion of this out-put (4%) was deflected to a photodetector (DET10A, Thorlabs, Newton, NJ)to compensate for pulse-to-pulse energy fluctuations and to provide optical trig-gering. A maximum pulse energy of 6 mJ was used in this study was deliveredfrom the distal end of the fiber (flat-cleaved at normal incidence). Control of PAimage acquisition was realized using a logic AND gate with two inputs: the opti-cal trigger and a digital control window provided by a LabView control programvia a digital I/O card (NI-USB-6501, National Instruments, Berkshire, UK). PA

374 W. Xia et al.

image reconstruction was performed in real-time using a custom delay-and-sumbeam-forming algorithm; offline, a more accurate Fourier-domain reconstructionalgorithm [7] was used.

Ultrasonic Tracking. A fiber-optic hydrophone (125 µm cladding, PrecisionAcoustics, Dorchester, UK) [8] was used to receive US transmissions from thelinear array probe (Figure 1a). Each element of the linear array was excitedby a bipolar electrical pulse (Figure 2a). The fiber-optic hydrophone (FOH)data collection were synchronized with the US tracking transmissions using twooutput triggers: a frame trigger corresponding to the start of each image frame,and a line trigger corresponding to the start of each transmission. Using the frametrigger, the line trigger signal and the FOH data was digitized using a DAQ card(USB-5132, National Instrument, Austin, Texas) as illustrated in Figure 1a. Theline triggers were used to parse the FOH data according to the start of each UStracking transmission. With knowledge of the sound speed in the medium, thetime-of-flights of the US pulses determine the distance between the hydrophonetip and the centers of the transmitting elements (Figure 2a). The peaks of thetime-gated signals form a parabolic shape (Figure 2b). A received transmissionfrom transducer element number 30 is shown in Figure 2c. An image of thehydrophone tip was reconstructed by applying a Fourier-domain reconstructionalgorithm which was developed initially for PA imaging [7] (Figure 2d). Animportant advantage of the ultrasonic tracking images is that they are inherentlyco-registered with the conventional B-mode US images and the PA images.

The fiber-optic hydrophone and the excitation light delivery fiber were in-serted into the cannula of a 14 gauge spinal needle (Terumo, Surrey, UK), whichserved as a surrogate for the fetoscope. The distal ends of the fibers were flushwith the bevel surface of the needle (Figure 1a).

2.2 Imaging and Tracking with a Human Placenta

A term human placenta was collected after a caesarean section delivery at theUniversity College London Hospital (UCLH). The study was approved by aJoint Committees of UCL and UCLH on the Ethics of Human Research and theplacenta was collected after written informed consent from the mother. Followingdelivery, the umbilical cord was clamped immediately to preserve the maximumamount of blood inside the placental fetal vasculature.

PA imaging of the placenta was performed with a block of gel positionedbetween the US probe and the fetal surface of the placenta. This gel, whichwas optically and acoustically transparent, served to simulate the amniotic fluidwithin the gestation sac (Figure 1b). During PA imaging, it was chosen in place ofsaline or similar aqueous solutions to limit diffusion of blood out of the placenta.With ultrasonic tracking, gel was not used as its acoustic coupling with thehydrophone was poor; the placenta was immersed in a saline and the fetoscopesurrogate was inserted to the placenta surface at an angle of 62 degrees (measuredrelative to the US probe surface).

Interventional Photoacoustic Imaging of the Human Placenta 375

Fig. 2. (a) Schematic illustration of single-element excitation of the linear array imag-ing probe to acquire a tracking image. (b) Time-gated transmissions received by thehydrophone. (c) The received transmission from transducer element number 30. (d)The reconstructed tracking image.

3 Results and Discussion

With PA imaging, a portion of the placental surface that was illuminated withexcitation light was visualized (Figure 3a-c). A circular region with high sig-nal amplitudes, which corresponded to a major vein on the placental surface,featured prominently in the PA images. A decrease in the average PA signalamplitude across the wavelength range of 760 nm to 840 nm (Figure 3d) wasalso apparent in the images (Figure 3a-c). Across the wavelength range of 750to 900 nm, good correspondence between the average PA signal and the opticalabsorption spectrum of deoxygenated blood was observed (Figure 4d). Absorb-ing structures at a depth in tissue of approximately 5 mm were apparent, whichwere attributed to small blood vessels. The appearance of the fetoscope surro-gate (needle), which is consistent with previous studies [9,10], can be attributedto photoacoustic excitation from back-scattered excitation light. The PA signalfrom the fiber could be used to provide an indication of the fiber location rela-tive to B-mode US images; however, severe artifacts resulting from US reflectionswithin the fetoscope surrogate are likely to limit the accuracy of this method.

With ultrasonic tracking, the hydrophone tip was clearly visualized (Figure 4).At two depths (25 mm and 41 mm), the US (Figure 4a,d) and tracking (Figure4b,e) images of the tip of the fetoscope surrogate had excellent spatial agreement.High SNRs were achieved for the tracking images (490 and 480, respectively).The full width at half maximum values of the axial and lateral profiles, which

376 W. Xia et al.

Fig. 3. Photoacoustic images of the placenta at wavelengths of (a) 750 nm, (b) 800nm and (c) 850 nm. The averaged photoacoustic amplitude over the region of interest(white box) as a function of excitation light wavelengths are shown in (d), with the op-tical absorption spectra of oxygenated and deoxygenated blood scaled to the averagedphotoacoustic amplitude at 750 nm.

Fig. 4. The hydrophone tip tracked at two locations within the ultrasound image plane.Two ultrasound images are shown in (a) and (d) with corresponding tracking imagesshown in (b) (e) respectively. The axial and lateral profiles for the hydrophone tipimage at two locations are shown in (c) and (f) with the full width at half maximumvalues indicating the accuracy of the tracking method.

Interventional Photoacoustic Imaging of the Human Placenta 377

can be taken as measures of the tracking accuracy, were consistently in a sub-millimeter level (Figure 4c,f).

In the context of minimally invasive fetal surgery, there are several potentialbenefits of ultrasonic tracking. First, knowledge of the position of the fiber thatdelivers excitation light relative to the US imaging probe could be used to modelthe light and US propagation, and further to facilitate the accurate recovery ofoptical absorption spectra of chromophores. Ultrasonic tracking of the fetoscopetip, which would be possible if the hydrophone were integrated directly to the fe-toscope, could facilitate fusion of fetoscopic and US images, as well as mosaicing offetoscopic images. Furthermore, the orientation of the fetoscope relative to the USprobe could be tracked using multiple hydrophone sensors distributed along it.

In the context of medical device tracking, an optical hydrophone as an USsensor has several advantages relative to a conventional piezoelectric US trans-ducer [8,11]. First, it is narrow and flexible, which facilitates integration intodevices. Second, it provides sufficient sensitivity for tracking, as demonstratedwith a high SNR (Figure 4). Finally, it possesses nearly omnidirectional sensi-tivity below 25 MHz, which allows the tracking with steep insertion angles.

The experimental paradigm in this study has several limitations. First, whilethe gel block phantom was designed to represent typical experiences encounteredin clinical practice, it did not allow for very large distances between the US imag-ing probe and the placenta such as those that may be encountered with obesepatients. Second, it would have been useful to have co-registration with othermodalities such as MRI and CT, but the accuracy would be limited by placentaldeformations. Third, follow-on studies are required to thoroughly compare thetracking method used in this paper with others. An alternative tracking methodto consider is the use of an US reflector on the fetoscope.

Interventional photoacoustic imaging has attracted great attention in recentyears [10,12,13,14,15]. However, to the authors knowledge, this proof-of-conceptstudy is the first to provide multispectral photoacoustic imaging of the humanplacenta, and also the first to provide ultrasonic tracking in a fetoscopic context.It sets the stage for comprehensive PA imaging across the surface of the placentain both normal and pathological cases in concert with histology. In future stud-ies, perfusion models that incorporate mechanisms for modifying the blood oxy-genation could be used to assess the sensitivity of PA images to blood flow [16]and oxygen saturation [5]. Further multispectral PA imaging could be used to dis-criminate between coagulated and non-coagulated blood during the treatment ofTTTS, based on their absorption spectra [17]. Photoacoustic imaging has strongpotential to improve visualization of placental vasculature during minimally in-vasive treatment of TTTS, and thereby to improve procedural outcomes.

Acknowledgments. This work was supported by an Innovative Engineer-ing for Health award by the Wellcome Trust [WT101957] and the Engineer-ing and Physical Sciences Research Council (EPSRC) [NS/A000027/1], by aStarting Grant from the European Research Council [ERC-2012-StG, Proposal310970 MOPHIM], and by an EPSRC First Grant [EP/J010952/1]. The authors

378 W. Xia et al.

acknowledge support from the Biomedical Research Centre of the United King-dom National Institute for Health Research (NIHR).

References

1. Slaghekke, F., et al.: Fetoscopic laser coagulation of the vascular equator versusselective coagulation for twin-to-twin transfusion syndrome: an open-label ran-domised controlled trial. Lancet 383, 2144–2151 (2014)

2. Pretorius, D.H., et al.: Imaging of placental vasculature using three-dimensionalultrasound and color power Doppler: a preliminary study. Ultrasound Obstet. Gy-necol. 21(1), 45–49 (1998)

3. Noguchi, J., et al.: Placental vascular sonobiopsy using three-dimensional powerDoppler ultrasound in normal and growth restricted fetuses. Placenta 30(5),391–397 (2009)

4. Beard, P.C.: Biomedical photocoustic imaging. Interface Focus 1(4), 602–631 (2011)5. Cox, B., et al.: Quantitative spectroscopic photoacoustic imaging: a review. J.

Biomed. Opt. 17(6), 061202 (2012)6. Xia, W., et al.: An optimized ultrasound detector for photoacoustic breast tomog-

raphy. Med. Phys. 40(3), 032901 (2013)7. Treeby, B.E., Cox, B.T.: k-Wave: MATLAB toolbox for the simulation and recon-

struction of photoacoustic wave fields. J. Biomed. Opt. 15(2), 021314 (2010)8. Morris, P., et al.: A Fabry-Perot fiber-optic ultrasonic hydrophone for the si-

multaneous measurement of temperature and acoustic pressure. J. Acoust. Soc.Am. 125(6), 3611–3622 (2009)

9. Su, J., et al.: Photoacoustic imaging of clinical metal needles in tissue. J. Biomed.Opt. 15(2), 021309 (2010)

10. Piras, D., et al.: Photoacoustic needle: minimally invasive guidance to biopsy. J.Biomed. Opt. 18(7), 070502 (2013)

11. Mung, J., Vignon, F., Jain, A.: A non-disruptive technology for robust 3D tooltracking for ultrasound-guided interventions. In: Fichtinger, G., Martel, A., Peters,T., et al. (eds.) MICCAI 2011, Part I. LNCS, vol. 6891, pp. 153–160. Springer,Heidelberg (2011)

12. Lediju Bell, M.A., et al.: in vivo visualization of prostate brachytherapy seeds withphotoacoustic imaging. J. Biomed. Opt. 19(12), 126011 (2014)

13. Lediju Bell, M.A., et al.: Transurethral light delivery for prostate photoacousticimaging. J. Biomed. Opt. 20(3), 036002 (2015)

14. Tavakoli, B., et al.: Detecting occlusion inside a ventricular catheter using photoa-coustic imaging through skull. In: Proc. SPIE 8943, p. 89434O (2014)

15. Lin, L., et al.: in vivo deep brain imaging of rats using oral-cavity illuminatedphotoacoustic computed tomography. J. Biomed. Opt. 20(1), 016019 (2015)

16. Brunker, J., Beard, P.: Pulsed photoacoustic Doppler flowmetry using time-domaincross-correlation: Accuracy, resolution and scalability. J. Acoust. Soc. Am. 132(3),1780–1791 (2012)

17. Talbert, R., et al.: Photoaoustic discrimination of viable and thermally coagulatedblood using a two-wavelength method for burn injury monitoring. Phys. Med.Biol. 52, 1815–1829 (2007)